The Use of Least-Squares for XPS Peak Parameters Estimation: Parts1. Myths and Realities

Statistical tools and diagnostics readily available in regression analysis could prove very useful for the estimation of XPS peak parameters. We describe here how XPS regression software should be written in order to include easy-to-interpret diagnostics enabling the user better to judge the quality of the regression. The paper is aimed both at the programmer who wishes to write his or her own software and at the eventual end-user who wishes only to interpret the diagnostics. Some drawbacks of currently available commercial software are discussed. Forthcoming papers will address the problems of interpretation of regression estimates, hypothesis testing and ill-conditioning.